Quite simple...they are incompatible traits. The pastel gene overrides the same wild-type alleles. This does not make the wild-type gene recessive. Of course, for each trait for which we've found a morph, there is a "wild-type" or "normal" gene. When we say an animal is "normal", we assume that the genes affecting ALL characteristics are normal. There would be a separate "normal" gene for alleles related to melanin production (T- albinism, ghosts/hypos), tyrosinase (T /T- Albinism), as well as such things as pattern (although the extent to which genetics controls different pattern types is still debatable), size, body shape, we could probably go on and on for days.
I don't know if anyone has actually determined, yet, what the gene for a Pastel actually controls. I'm still relatively new to BP's, so if anyone knows of the pigmentation characterics of Pastels I'd appreciate some insight there.
As genetics is based mainly on theory (which is still very similar to Mendel's early works on the subject), it is impossible for anyone to be sure exactly how this all works. But these theories seem to hold up well under nearly all scrutiny that's been thrown at them. I've heard of a group of scientists who contend that there is no such thing as "recessive" genes, and that those traits we consider to be "simple recessive" are actually traits that are only observed in a co-dominant form. The question is purely academic, as the actual practice of breeding desirable traits would not be affected much, but it is interesting to ponder once in a while.